Method, system and computer program product for pre-classifying drugs based on drug characteristics to reduce artificial intelligence model size for validating drugs

Abstract

A method comprising: receiving information associated with a medication, the information comprising a plurality of characteristics; filtering the information based on at least one of the plurality of characteristics to identify one of the plurality of artificial intelligence engines; and predicting a national drug code (NDC) of the drug using the identified one of the plurality of artificial intelligence engines, or verifying whether the drug matches the target drug using the identified one of the plurality of artificial intelligence engines.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of U.S. Provisional Application No. 63 / 366,701, filed on June 21, 2022, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0002] The present disclosure relates generally to identification of pharmaceuticals, and in particular, to methods, systems, and computer program products for identifying pharmaceuticals, for example to facilitate verification of the contents of a pharmaceutical packaging. Background Art

[0003] Drug packaging systems can be used in facilities such as pharmacies, hospitals, long-term care facilities, etc. to dispense medications according to prescriptions. These drug packaging systems can include systems designed to package medications in various container types (including but not limited to bags, vials, bottles, blister cards, and strip packs). Strip packs are a type of packaging in which medications are packaged in individual bags for administration on a specific date (and in some cases, at a specific time). Typically, the individual bags are detachably connected together and are typically provided in the form of rolls. When necessary, the bags can be separated from the roll. Before the medication is issued to the customer, the contents of the medication package can be verified to ensure that the customer receives the correct medication. Summary of the invention

[0004] In some embodiments of the present inventive concept, a method includes: receiving information associated with a drug, the information including multiple characteristics; filtering the information based on at least one of the multiple characteristics to identify one of a plurality of artificial intelligence engines; and using the identified one of the multiple artificial intelligence engines to predict a national drug code (NDC) of the drug, or using the identified one of the multiple artificial intelligence engines to verify whether the drug matches a target drug.

[0005] In other embodiments, the plurality of artificial intelligence engines respectively correspond to a plurality of value combinations of at least one characteristic among the plurality of characteristics.

[0006] In other embodiments, at least one of the plurality of characteristics is not used to filter information. At least one of the plurality of characteristics that is not used to filter information is used as at least one feature for training the plurality of artificial intelligence engines.

[0007] In other embodiments, the plurality of characteristics includes size, shape, color, imprint code, or score.

[0008] In other embodiments, the imprint code includes indicia of pharmaceutical strength, indicia of active ingredients, and indicia of inactive ingredients.

[0009] In other embodiments, the shapes include circles, ovals, and other shapes.

[0010] In other embodiments, the color includes clear and multiple colors.

[0011] In other embodiments, filtering the information includes filtering the information based on all of a plurality of characteristics to identify an artificial intelligence engine from a plurality of artificial intelligence engines.

[0012] In other embodiments, each of the plurality of artificial intelligence engines comprises a convolutional neural network.

[0013] In other embodiments, the convolutional neural network includes a plurality of convolutional layers, wherein at least some of the plurality of convolutional layers are connected to each other via skip connections.

[0014] In other embodiments, the plurality of artificial intelligence engines respectively correspond to a plurality of value combinations of at least one of the plurality of characteristics. The method further includes granting the entity access to some of the plurality of artificial intelligence engines based on the entity distributing some of the plurality of drugs having some of the plurality of value combinations corresponding to some of the plurality of artificial intelligence engines.

[0015] In some embodiments, a method includes: receiving training information associated with a drug from multiple sources, the training information including multiple characteristics and a national drug code (NDC) of the drug; determining whether to accept, reject, or waitlist the training information associated with the drug based on consistency of the training information between some of the multiple sources; and when accepting the training information, using the training information to train an artificial intelligence engine configured to predict the NDC code of the drug based on the multiple characteristics.

[0016] In other embodiments, determining whether to accept, reject, or waitlist the training information includes accepting the training information when the training information is consistent between at least a consensus subset of ones of the plurality of sources.

[0017] In other embodiments, the consensus subset includes a minimum number X of the plurality of sources.

[0018] In other embodiments, the consensus subset also includes a minimum percentage Y of the plurality of sources.

[0019] In other embodiments, determining whether to accept, reject, or waitlist the training information includes using a consensus artificial intelligence engine to determine whether to accept, reject, or waitlist the training information.

[0020] In yet other embodiments, the consensus artificial intelligence engine uses K-means clustering to determine whether to accept, reject, or wait-list the training information.

[0021] In yet other embodiments, the method further includes: prior to accepting the training information, validating the training information with a trusted source of the training information.

[0022] In yet other embodiments, the plurality of characteristics includes size, shape, color, imprint code, or rating.

[0023] In some embodiments, a system includes: a processor; and a memory, coupled to the processor and including computer-readable program code embodied in the memory and executable by the processor to perform operations, the operations including: receiving information associated with a drug, the information including multiple characteristics; filtering the information based on at least one of the multiple characteristics to identify one of multiple artificial intelligence engines; and using the identified one of the multiple artificial intelligence engines to predict a national drug code (NDC) of the drug, or using the identified one of the multiple artificial intelligence engines to verify whether the drug matches a target drug.

[0024] In some embodiments, a computer program product includes: a non-transitory computer-readable storage medium, including a computer-readable program code embodied in the medium and executable by a processor to perform operations, the operations including: receiving information associated with a drug, the information including multiple characteristics; filtering the information based on at least one of the multiple characteristics to identify one of multiple artificial intelligence engines; and using the identified one of the multiple artificial intelligence engines to predict a national drug code (NDC) of the drug, or using the identified one of the multiple artificial intelligence engines to verify whether the drug matches a target drug.

[0025] In some embodiments, a system includes: a processor; and a memory, coupled to the processor and including computer-readable program code embodied in the memory and executable by the processor to perform operations, the operations including: receiving training information associated with a drug from multiple sources, the training information including multiple characteristics and a national drug code (NDC) of the drug; determining to accept, reject, or place the training information associated with the drug on a wait list based on consistency between some of the multiple sources of the training information; and when accepting the training information, using the training information to train an artificial intelligence engine configured to predict the NDC code of the drug based on the multiple characteristics.

[0026] In some embodiments, a computer program product includes: a non-transitory computer-readable storage medium, including a computer-readable program code embodied in the medium and executable by a processor to perform operations, the operations including: receiving training information associated with a drug from multiple sources, the training information including multiple characteristics and a national drug code (NDC) of the drug; determining whether to accept, reject, or put the training information associated with the drug on a waiting list based on consistency between some of the multiple sources of the training information; and when accepting the training information, using the training information to train an artificial intelligence engine, the artificial intelligence engine configured to predict the NDC code of the drug based on the multiple characteristics.

[0027] After reviewing the following drawings and detailed description, other methods, systems, articles of manufacture and / or computer program products according to embodiments of the present invention will be or become apparent to those skilled in the art. It is intended that all such additional systems, methods, articles of manufacture and / or computer program products be included in this description, within the scope of the present invention, and protected by the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Other features of the embodiments will be more readily understood from the following detailed description of specific embodiments thereof when read in conjunction with the accompanying drawings, in which:

[0029] Figure 1 is a block diagram illustrating a communication network including an artificial intelligence (AI) assisted drug analysis system according to some embodiments of the inventive concept;

[0030] Figure 2 According to some embodiments of the present invention Figure 1 Block diagram of the AI-assisted drug analysis system;

[0031] Figure 3 is a block diagram of a convolutional neural network for predicting a national drug code (NDC) of a drug according to some embodiments of the inventive concept;

[0032] Figure 4 According to some embodiments of the present invention Figure 3 Block diagram of the skip connection arrangement between convolutional layers of a convolutional neural network;

[0033] Figure 5 is a block diagram illustrating preprocessing of drug images according to some embodiments of the present inventive concept;

[0034] Figure 6 is a block diagram illustrating pre-filtering or pre-classifying drug information based on characteristics or features according to some embodiments of the present inventive concept;

[0035] Figure 7is a flow chart illustrating operations for performing drug analysis according to some embodiments of the present inventive concept;

[0036] Figure 8 is a block diagram illustrating a consensus system for verifying drug training information for training an AI system according to some embodiments of the present inventive concept;

[0037] Fig. 9 is a flowchart illustrating operations for verifying drug training information for training an AI system according to some embodiments of the present inventive concept;

[0038] Fig.10 According to some embodiments of the present invention, it can be used to implement Figure 1 A data processing system for one or more servers in an AI-assisted drug analysis system; and

[0039] Fig.11 is a diagram showing some embodiments of the present invention. Figure 1 Block diagram of the software / hardware architecture of the AI-assisted drug analysis system. DETAILED DESCRIPTION

[0040] In the following detailed description, many specific details are set forth to fully understand the embodiments of the inventive concept. However, it will be appreciated by those skilled in the art that the present invention can be practiced without these specific details. In some cases, well-known methods, processes, components, and circuits are not described in detail to avoid obscuring the inventive concept. It is intended that all embodiments disclosed herein can be implemented individually or combined in any manner and / or combination. The various aspects described for an embodiment can be incorporated into different embodiments, although not specifically described in this regard. That is, all embodiments and / or the features of any embodiment can be combined in any manner and / or combination.

[0041] As used herein, the term "data processing facility" includes, but is not limited to, hardware elements, firmware components, and / or software components. A data processing system may be configured with one or more data processing facilities.

[0042] As used herein, the term "drug packaging system" refers to any type of drug dispensing system, including but not limited to automated systems that fill vials, bottles, containers, pouches, blister cards, etc. with drugs, semi-automated systems that fill vials, bottles, containers, pouches, blister cards, etc. with drugs, and any combination of automated and semi-automated systems for filling drug packages with drugs. Drug packaging systems also include packaging systems for drug substitutes such as nutritional supplements and / or biologics.

[0043] As used herein, the terms "drug" and "medicine" are interchangeable and refer to a drug prescribed to a human or animal patient. A drug or medicine can be embodied in a variety of ways, including but not limited to pill form, capsule form, tablet form, and the like.

[0044] The term "drug" refers to any type of medicine that can be packaged in vials, bottles, containers, pouches, blister cards, etc. by automatic and semi-automatic drug packaging systems, including but not limited to pills, capsules, tablets, caplets, gel caps, lozenges, etc. Drugs also refer to drug substitutes such as nutritional supplements and / or biologics. An example drug packaging system, including management technology for completing packaging orders, is described in U.S. Pat. No. 10,492,987, the disclosure of which is incorporated herein by reference.

[0045] The term "drug package" refers to any type of object that can hold a drug, including but not limited to vials, bottles, containers, pouches, blister cards, and the like.

[0046] Embodiments of the inventive concept are described herein in the context of a drug analysis engine including one or more machine learning engines and artificial intelligence (AI) engines. It will be understood that embodiments of the inventive concept are not limited to specific implementations of the drug analysis engine, and various types of AI systems may be used, including but not limited to multi-layer neural networks, deep learning systems, natural language processing systems, and / or computer vision systems. In addition, it will be understood that a multi-layer neural network is a multi-layer artificial neural network including artificial neurons or nodes, and does not include a biological neural network including real biological neurons. Embodiments of the inventive concept may be implemented using multiple AI systems, or may be implemented by combining various functions into fewer or single AI systems. The AI ​​engine described herein may be configured to convert the memory of a computer system to include one or more data structures, such as but not limited to arrays, scalable arrays, linked lists, binary trees, balanced trees, heaps, stacks, and / or queues. When a computer system operates in an inference mode to perform inference, prediction, classification, suggestions, etc. in response to input information or data provided thereto, these data structures may be configured or modified by an AI training process to improve the efficiency of the computer system.

[0047] When a drug packaging system is used to package drugs for delivery to customers, a verification or audit process may be performed to ensure that the packaged drugs correspond to the patient's prescription. In some cases, the drug packaging entity may use an artificial intelligence (AI) system to analyze drug images and / or other drug information to predict the corresponding national drug code (NDC), thereby determining whether the NDC code of the drug matches the NDC of the drug identified in the prescription. Typically, the AI ​​system uses a single larger overall model that is trained using drug images and / or other drug information of a large number of drugs that can be distributed across a large number of packaging entities (i.e., different pharmacies, hospitals, etc.). However, as new drugs are produced, retraining a larger AI model may consume a lot of time and resources. In addition, many drug distribution entities may only distribute a small subset of the total number of drugs used to train a larger AI model. Therefore, some embodiments of the present invention are conceived from such a recognition that an AI system including an AI model for training a large number of drugs may be inefficient in verifying or auditing packaged drugs. In addition, many distribution entities may not distribute all or even many drugs used to train the AI ​​model. Thus, the benefits of an AI model being able to recognize a more comprehensive list of medications may be less useful to an individual drug dispensing entity that dispenses a relatively small subset of medications and may prefer more flexible and efficient training of AI systems and models targeted to those medications dispensed by the dispensing entity.

[0048] Accordingly, some embodiments of the present invention may provide a drug analysis engine that may be configured to receive information associated with a drug including one or more characteristics of the drug, and to pre-classify or pre-filter the information based on one or more characteristics of the drug. Pre-filtering or pre-classifying the drug based on one or more of the characteristics of the drug may allow the drug to be classified or grouped. Multiple AI engines or models corresponding to different drug groups generated based on pre-classification or pre-filtering may be developed. In order to predict the NDC of a drug or verify whether a drug matches a target drug, an AI engine or model corresponding to the group into which the drug falls based on pre-classification or pre-filtering may be selected, and the drug information of the drug may be provided to the selected AI engine or model. The AI ​​engine or model may then predict the NDC code and / or verify whether the drug matches the target drug based on its training based on the information associated with the drug.

[0049] Therefore, instead of using a single larger AI engine or model, multiple AI engines or models that are customized for subsets of drugs classified based on one or more of the drug characteristics can be used. Therefore, a drug distribution entity can subscribe to access only the AI ​​engines or models corresponding to the drugs it distributes. In addition, drug characteristics that are not used in pre-classification or filtering can be used as features for training different AI engines or models. Because the model is only trained for a subset of drugs based on all possible combinations of drug characteristics, and fewer features are used during the training process, the training of separate AI engines or models can be more efficient. NDC prediction systems and / or drug matching verification systems using distributed models can also be more efficient. Increased data throughput can enhance system performance and help address the challenges of frequent model training.

[0050] As described above, information used to train an AI engine or model (which may include data labels, NDC numbers, etc.) is typically provided by a drug distributing entity (e.g., a pharmacy, a hospital, etc.). For example, when a new drug is introduced, the drug distributing entity may provide training information associated with the new drug, the training information including one or more images of the drug, one or more characteristics of the drug, metadata associated with the drug, and / or the NDC of the drug. At least some of this training information is typically entered manually and may be prone to errors. Some embodiments of the present inventive concept may provide a consensus engine that is configured to receive training information associated with a drug and is configured to determine whether to accept the training information for training one or more AI engines or models, to list the training information on a waiting list as a candidate for use in training, or to reject the training information. For example, the consensus engine may use rules and thresholds to allow acceptance of training information once a consensus is reached on the number or percentage of sources that provide the same training information for a drug. In other embodiments, the consensus engine may be implemented as an AI system or model trained based on historical drug information from various sources to learn when the training information can be accepted as valid, when the training information should be placed on a waiting list for acceptance, and when the training information should be rejected. The consensus engine (which may be an AI machine learning system) may use votes from various sources as described above, and / or may also use trusted sources (e.g., trusted reference books or manuals, and / or trusted image libraries or databases) that can be used to confirm training information including NDCs for accepting AI systems or models used in training NDCs that predict drugs, and / or for verifying whether a drug matches a target drug.

[0051] Reference Figure 1According to some embodiments of the present inventive concept, as shown in the figure, a communication network 100 including an AI-assisted drug analysis system includes a pharmacy management system (PMS) or a host system 110 coupled via a network 140, a packaging system server 120, a drug analysis engine server 155, and one or more drug packaging systems 130a and 130b.

[0052] The PMS system 110 can be configured to manage and fill prescriptions for customers. As used herein, the PMS system can be used in a pharmacy, or can generally be used as a batch generation system for other applications (such as dispensing nutrition or biological products). The PMS system 110 can be associated with various types of facilities (such as pharmacies, hospitals, long-term care facilities, etc.). The PMS system or host system 110 can be a system that can send a valid prescription to one or more drug packaging systems 130a and 130b. The packaging system server 120 may include a packaging system interface module 135, and may be configured to manage the operation of drug packaging systems 130a and 130b. For example, the packaging system server 120 may be configured to receive a packaging order from the PMS system 110, and identify which of the drug packaging systems 130a and 130b should be used to package a specific single order or a specific batch order. In addition, the packaging system server 120 may be configured to manage the operation of drug packaging systems 130a and 130b. For example, the packaging system server 120 may be configured to manage the inventory of medications available through each of the medication packaging systems 130a and 130b, manage medication dispensing canisters assigned to or registered with one or more of the medication packaging systems 130a and 130b, manage the general operational status of the medication packaging systems 130a and 130b, and / or manage reports related to the status (e.g., assigned, completed, etc.) of packaging orders, medication inventory, order billing, etc. A user 150 (such as a pharmacist or pharmacy technician) may communicate with the packaging system server 120 using any suitable computing device via a wired and / or wireless connection. Although Figure 1 140, it will be appreciated that the user 150 may communicate with the packaging system server 120 via one or more network connections (e.g., via the network 140). The user 150 may interact with the packaging system server 120 to approve or override various recommendations made by the packaging system server 120 in operating the drug packaging systems 130a and 130b. The user 150 may also initiate the running of various reports for the drug packaging systems 130a and 130b as described above. Although Figure 1 Only two drug packaging systems 130a and 130b are shown in FIG. 1 , but it will be understood that the packaging system server 120 may manage more than two drug packaging systems.

[0053] The AI-assisted drug analysis system may include a drug analysis engine server 155, which includes a drug analysis engine module 160, which is configured to facilitate, for example, verification of packaged and / or unpackaged drugs. The drug analysis engine server 155 and the drug analysis engine module 160 may represent one or more AI systems that are trained and operated in reasoning mode by pre-filtering or pre-classifying drugs based on one or more characteristics of the drugs (which may allow the drugs to be classified or grouped). Therefore, the drug analysis engine server 155 may represent one or more AI engines or models corresponding to the drug categories or groups identified by pre-filtering or pre-classification, respectively. As described above, as new drugs are developed and added to the system, smaller AI engines or models corresponding to the categories or groups of various drugs may be more efficiently trained, and the smaller AI engine or model may be more relevant to a separate drug distribution entity that may only distribute a subset of drugs to its customers.

[0054] It will be understood that the division of functionality between the packaging system server 120 / packaging system interface module 135 and the drug analysis engine server 155 / drug analysis engine module 160 described herein is merely exemplary. According to different embodiments of the inventive concept, various functions and capabilities may be moved between the packaging system server 120 / packaging system interface module 135 and the drug analysis engine server 155 / drug analysis engine module 160. Furthermore, in some embodiments, the packaging system server 120 / packaging system interface module 135 and the drug analysis engine server 155 / drug analysis engine module 160 may be merged into a single logical and / or physical entity.

[0055] The network 140 couples the drug packaging systems 130a and 130b, the PMS system 110, the packaging system server 120, and the drug analysis engine server 155 to each other. The network 140 may be a global network such as the Internet or other publicly accessible networks. The various elements of the network 140 may be interconnected through a wide area network, a local area network, an intranet, and / or other private networks that may not be accessible to the public. Therefore, the network 140 may be a communication network, and / or may represent a combination of public and private networks, or a virtual private network (VPN). The network 140 may be a wireless network, a wired network, or may be a combination of both wireless and wired networks. In some embodiments, the drug analysis engine server 155 may also be coupled to the network 140.

[0056] In some embodiments, the AI-assisted drug analysis service provided by the drug analysis engine server 155 and the drug analysis engine module 160 may be implemented as a cloud service. In some embodiments, the AI-assisted drug analysis service may be implemented as a representational state transfer web service (RESTful Web service).

[0057] although Figure 1 An example communication network including an AI-assisted drug analysis system is shown, but it will be understood that embodiments of the present subject matter are not limited to such configurations, but are intended to encompass any configuration capable of performing the operations described herein.

[0058] As described above, drug analysis engine server 155 and drug analysis engine module 160 may represent one or more AI systems that may, for example, facilitate verification of packaged and / or unpackaged drugs. Figure 2 is a block diagram of a drug analysis engine module 160 embodied as an AI system, which includes one or more AI engines or models (such as machine learning systems) that can be used to detect packaged and / or unpackaged drugs, and / or identify these drugs that have been detected through NDC. Figure 2 The AI ​​system of can represent a single AI engine or model that can be used to identify drugs corresponding to a single group based on pre-filtering or pre-classifying the drugs according to one or more characteristics of the drugs. Figure 2 The architecture of the AI ​​system of the invention can be replicated to form a separate AI system to detect packaged and / or unpackaged drugs respectively and identify these drugs that have been detected by NDC. Figure 2 As shown, the drug analysis engine module 160 may include a training module and a module for processing new data to detect and / or identify packaged and / or unpackaged drugs in an image. The modules used in the training portion of the drug analysis engine module 160 may include a training data module 205, a characterization module 225, a labeling module 230, and a machine learning engine 240.

[0059] The training data module 205 may be configured to obtain and / or store training data, which may include one or more images of packaged and / or unpackaged drugs, and additional information or data associated with each drug (such as characteristics of the drug). The training data stored by the training data module 205 may also include the NDC of each drug. Figure 2 The machine learning architecture is shown in , but other embodiments may use Figure 2 An artificial neural network other than an embodiment of a machine learning system of the invention, or using an artificial neural network instead of Figure 2Embodiment of a machine learning system. The characterization module 225 is configured to identify separate independent variables (which can be considered dependent variables) used by the drug analysis engine module 160 to detect and / or identify one or more drugs in images of packaged and / or unpackaged drugs, for example. The training data may generally not be processed or formatted, and the training data may also include additional information in addition to the drug and / or drug packaging information. For example, the training data may include account codes, business address information, etc. that may be filtered out by the characterization module 225. Features extracted from the training data may be referred to as attributes, and the number of features may be referred to as dimensions. According to some embodiments of the present invention, one or more characteristics of a drug may be used for pre-filtering or pre-classification of the drug as will be described below. Since it is known that the drug for which the AI ​​engine or model is being trained has these one or more characteristics, these characteristics do not have to be feature candidates identified by the characterization module 225.

[0060] The labeling module 230 can be configured to assign defined labels to the training data and the detected and / or identified drugs to ensure that the naming conventions for both the input characteristics and the generated outputs are consistent. The machine learning engine 240 can process the characterized training data including the labels provided by the labeling module 230, and can be configured to test multiple functions to establish a quantitative relationship between the characterized and labeled input data and the generated outputs. The machine learning engine 240 can use modeling techniques to evaluate the impact of various input data features on the generated outputs. These impacts can then be used to tune and refine the quantitative relationship between the characterized and labeled input data and the generated outputs. The tuned and refined quantitative relationship between the characterized and labeled input data generated by the machine learning engine 240 is output for use in the AI ​​engine 245. The machine learning engine 240 can be referred to as a machine learning algorithm.

[0061] The modules for detecting packaged and / or unpackaged drugs, identifying these drugs that have been detected by NDC in the image, and / or verifying whether these drugs match one or more target drugs include a new data module 255, a characterization module 265, an AI engine 245, and a drug analysis module 275. The new data 255 can be data / information that is the same as the training data 205 in content and form, except that the new data 255 will be used to analyze new packaged and / or unpackaged drugs instead of for training purposes. Similarly, the functions performed by the characterization module 265 on the new data 255 are the same as the functions performed by the characterization module 225 on the training data 205. The AI ​​engine 245 can implement the effects generated by the machine learning engine 240 in the form of a quantitative relationship determined between the characterized and labeled input data and the output drug package content analysis. In some embodiments, the AI ​​engine 245 can be referred to as an AI model.

[0062] The AI ​​engine 245 may be configured to identify one or more drugs based on the NDC of the drug, and / or verify whether the drug matches the target drug. The AI ​​engine 245 may use various modeling techniques to detect packaged and / or unpackaged drugs and identify these drugs that have been detected in the image by NDC according to different embodiments of the present inventive concept, including but not limited to regression techniques, neural network techniques, autoregressive integrated moving average (ARIMA) techniques, deep learning techniques, linear discriminant analysis techniques, decision tree techniques, naive Bayes techniques, K-nearest neighbor techniques, learning vector quantization techniques, support vector machine techniques, and / or bagging / random forest techniques.

[0063] The drug analysis module 275 can be configured to output the NDC codes of packaged and / or unpackaged drug images with one or more drugs identified by one or more markers (such as bounding boxes) together with the NDC of the one or more drugs to a drug packaging verification system.

[0064] As described above, the drug analysis engine server 155 and the drug analysis engine module 160 may represent one or more AI systems that may be configured as the drug analysis engine server 155, and the drug analysis engine module 160 may represent one or more AI systems that may, for example, facilitate verification of packaged and / or unpackaged drugs. Figure 3 is a block diagram of a drug analysis engine module 160 for implementing an AI system by means of a neural network, which can be used to supplement and / or replace Figure 2 Machine learning embodiments of the invention can be used to detect packaged and / or unpackaged drugs, identify those drugs that have been detected by NDC, and / or verify that the drug matches a target drug. Figure 3 In an example embodiment of the invention, the neural network is a convolutional neural network. However, it will be understood that according to other embodiments of the present inventive concept, the AI ​​system for detecting packaged and / or unpackaged drugs, identifying such drugs that have been detected by NDC, and / or verifying whether a drug matches a target drug may also be embodied as a fully connected neural network. However, due to the large number of pixels and the resulting large number of weights that need to be managed in the neural network layers, convolutional neural networks can be useful in processing or classifying images. Convolutional neural networks can reduce the main image matrix to a matrix with lower dimensions through convolution in several layers, including hidden layers and activation layers, which reduces the number of weights used and reduces the impact on training time. The final layer can use a softmax function as the activation function in the output layer, which can predict a multinomial probability distribution that may match the NDC label.

[0065] Now refer to Figure 3, the image preprocessor 305 may receive one or more images of packaged and / or unpackaged medicines. Figure 6 As described above, the image preprocessor may perform various corrections on the image data, including, for example, gamma correction, noise reduction, and / or image segmentation. The preprocessed drug image may then be provided to the convolutional neural network 310, and the preprocessed drug image may be an image represented by a matrix of dimensions AxBx3, where the number 3 represents red, green, and blue. Figure 3 As shown, the convolutional neural network 310 includes a first convolutional layer 320 and a second convolutional layer 330, and a first pooling layer 325 and a second pooling layer 335. Each of the convolutional layers 320 and 330 is a matrix with a dimension smaller than the input matrix, and can be configured to perform a convolution operation using a portion of the input matrix with the same dimension. The sum of the products of the corresponding elements is the output of the convolutional layer. The output of each convolutional layer can also be processed by a rectified linear unit operation, in which any number below 0 is converted to 0 and any positive number remains unchanged. The convolutional neural network 310 also includes a first pooling layer 325 and a second pooling layer 335. The pooling layers 325 and 335 can each be configured to filter the outputs of the convolutional layers 320 and 330 by performing a downsampling operation, respectively. The size of the pooling operation or filter is smaller than the size of the feature map of the input. In some embodiments, the size of the pooling operation or filter is 2×2 pixels with a stride of 2 pixels applied. This means that the pooling layer will always reduce the size of each feature map by a factor of 2, e.g., each dimension is halved, reducing the number of pixels or values ​​in each feature map to one-quarter the size. For example, a pooling layer applied to a feature map of 6×6 (36 pixels) will result in an output pooled feature map of 3×3 (9 pixels). The final output layer is a normal fully connected neural network layer 340 that gives the output as the predicted drug NDC or drug verification 345.

[0066] In some embodiments of the present inventive concept, the convolutional neural network 310 may be a residual neural network using a skip connection between the convolutional layers 320 and 330 . Figure 4 An example of a skip connection is shown. Specifically, in a skip connection, a convolutional neural network involves a convolutional layer that receives both the output of a previous convolutional layer and the input of the previous convolutional layer as input.

[0067] It will be understood that although for illustrative purposes Figure 3 Two convolutional layers 320 and 330 are shown in the example convolutional neural network 310 of , but the convolutional neural network according to various embodiments of the inventive concept may include many convolutional layers, and in some embodiments may exceed 100 layers.

[0068] As described above, the pharmaceutical images may be preprocessed to perform various corrections on the image data. Figure 5 , the gamma correction module 505 can perform gamma correction on the drug image to generate a gamma-corrected image. One or more cameras can darken the image; gamma correction can brighten the image to allow the convolutional neural network 310 to better identify the edges of various elements displayed in the image. Except for low brightness, the gamma correction can be embodied as a power law transformation, which can be linear at low brightness to avoid having infinite derivatives at brightness zero. This is a traditional nonlinearity applied to encoding SDR images. The exponent or "gamma" can have a value of 0.45, but the linear portion of the lower part of the curve can make the final gamma correction function closer to a power law exponent of 0.5 (i.e., a square root transformation); therefore, the gamma correction can follow the DeVries-Rose law of brightness perception. The Gaussian blur denoising module 510 is used to perform Gaussian blur denoising on the gamma-corrected image to generate a denoised image. The Gaussian blur denoising module or filter 510 can be a linear filter. It can be used to blur the image and / or reduce noise. Two Gaussian blur denoising filters 510 may be used such that the outputs are subtracted for "unsharp masking" (edge ​​detection). The Gaussian blur denoising module or filter 510 may blur edges and reduce contrast. A median filter is a non-linear filter that may be used as a way to reduce noise in an image. An automatic image thresholding module 515 may perform automatic image thresholding on the denoised image to produce a foreground-background separated image. Thresholding is a technique used in image segmentation applications. Thresholding involves selecting a desired grayscale threshold for separating an object of interest in an image from the background based on the grayscale distribution. The Otsu method is a global thresholding method that depends only on the grayscale values ​​of the image. The Otsu method is a global thresholding selection method that involves calculating a grayscale histogram. When applied in only one dimension, it may not adequately segment the image. A two-dimensional Otsu method based on a grayscale threshold for each pixel and its spatially correlated information with the neighborhood around the pixel may be used. Therefore, the Otsu method may provide satisfactory segmentation when applied to noisy images. From Figure 5 The output image of the preprocessing module can be applied to the drug packaging modification engine (such as Figure 3 Convolutional Neural Network 310).

[0069] Figure 6 is a block diagram illustrating pre-filtering or pre-classifying drug information based on characteristics or features according to some embodiments of the present inventive concept. As described above, pre-filtering or pre-classifying drugs based on one or more characteristics of the drugs may allow the drugs to be classified or grouped. Figure 6 As shown, one or more filters or classifiers may be used to process drug information. Figure 6In the example of , N filters or classifiers are shown. The number of filters or classifiers can be based on the number of characteristics associated with the drug. In some embodiments, drug characteristics can include size, shape, color, imprint code, and / or score. Imprint code characteristics can include markings of medicinal strength, markings of active ingredients, and / or markings of inactive ingredients. Shape characteristics can be circular, elliptical, and / or other shapes. Color characteristics can be any of a variety of colors and / or transparent. Multiple AI engines or models (such as those described above for different groups of drugs generated based on pre-classification or pre-filtering) can be developed, each corresponding to a different group of drugs generated based on pre-classification or pre-filtering. Figure 2 and Figure 3 described in Figure 6 In the example shown, N AI engines are shown. If two characteristics are used that are each capable of assuming two unique values, four AI engines corresponding to all four value combinations of the two drug characteristics used in the pre-classification or pre-filtering process can be used. In order to predict the NDC of a drug, and / or verify whether the drug matches a target drug, an AI engine or model corresponding to the group into which the drug falls based on pre-classification or pre-filtering can be selected, and the drug information of the drug can be provided to the selected AI engine or model. The AI ​​engine or model can then predict the NDC code and / or verify whether the drug matches the target drug based on its training based on the information associated with the drug. Many drug distribution entities may only distribute Figure 6 A smaller subset of the total number of drugs covered by all N AI engines or models shown in . Thus, according to some embodiments of the present inventive concept, a drug dispensing entity may be granted access only to those AI engines or models associated with drug groups or categories covering drugs distributed by the drug dispensing entity. For example, access may be made via a cloud service, and / or an appropriate AI engine or model may be provided to the drug dispensing entity for execution on the drug dispensing entity's own platform.

[0070] Figure 7 is a flow chart illustrating operations for performing drug analysis according to some embodiments of the present inventive concept. Figure 7 , operations begin at block 700 where information associated with a drug is received. The information may include a plurality of drug characteristics, including but not limited to size, shape, color, imprint code, and / or rating. At block 705, the drug information is filtered or classified based on one or more of the plurality of characteristics to identify one of a plurality of AI engines or models. At block 710, the identified AI engine or model is used to predict the NDC of the drug and / or verify whether the drug matches a target drug.

[0071] As described above, the information used to train the AI ​​engine or model is typically provided by a drug dispensing entity (e.g., a pharmacy, hospital, etc.). When a new drug is introduced, the drug dispensing entity may provide training information associated with the new drug, the training information including one or more images of the drug, one or more characteristics of the drug (e.g., size, shape, color, imprint code, and / or rating), metadata associated with the drug, and / or the NDC of the drug. At least some of this training information is typically entered manually and may be prone to errors. Now referring to Fig. 9 The operation of the consensus engine 800 begins at block 900 where the consensus engine receives training information associated with a drug from a plurality of sources. Figure 8 As shown, three sources A 805a, source B 805b, and source C 805c can provide drug training information to the consensus engine 800. At block 905, the consensus engine can determine whether to accept the training information for use in training one or more AI engines or models, list the training information as a candidate for use in training, or reject the training information. At block 910, the accepted training information is used to train an AI engine that is configured to predict the NDC of a drug and / or verify whether the drug matches a target drug.

[0072] The consensus engine can be configured to use rules and thresholds, such as allowing training information to be accepted once the number or percentage of sources providing the same training information for a drug is consistent. For example, training information can be accepted when the training information is consistent between at least one consensus subset of some of the multiple sources. In some embodiments, the consensus subset includes a minimum number X of the multiple sources. In other embodiments, the consensus subset also includes a minimum percentage Y of the multiple sources.

[0073] In other embodiments, the consensus engine 800 can be implemented as an AI system or model trained based on historical drug information from various sources to learn when the training information can be accepted as valid, when the training information should be placed on a waiting list for acceptance, and when the training information should be rejected. In some embodiments, the consensus engine 800, when implemented using an AI engine or model, can use K-means clustering to determine whether to accept, reject, or place the training information on a waiting list.

[0074] The consensus engine 800 may also use trusted sources 810 (e.g., trusted reference books or manuals, and / or trusted image libraries or databases) that can be used to validate training information, including NDC, for use in accepting information for training an AI system or model used in predicting the NDC for a drug.

[0075] Now refer to Fig.10 According to some embodiments of the present invention, it can be used to implement Figure 1The data processing system 1000 of the drug analysis engine server 155 includes an input device 1002 (such as a keyboard or keypad, a barcode scanner, or an RFID reader), a display 1004, and a memory 1006 that communicates with a processor 1008. The data processing system 1000 may also include a storage system 1010, a speaker 1012, and an input / output (I / O) data port 1014 that also communicates with the processor 1008. The processor 1008 may be, for example, a commercially available or custom microprocessor. The storage system 1010 may include removable and / or fixed media (such as a floppy disk, a ZIP drive, a hard disk, etc.), and a virtual storage device (such as a RAMDISK). The I / O data port 1014 can be used to transmit information between the data processing system 1000 and another computer system or network (e.g., the Internet). These components may be conventional components (such as those used in many conventional computing devices), and their functions relative to conventional operations are generally known to those skilled in the art. According to some embodiments of the present inventive concepts, the memory 1006 may be configured with computer-readable program code 1016 to facilitate AI-assisted verification of packaged and / or unpackaged drugs.

[0076] According to some embodiments of the present invention, Fig.11 105 is shown, which can be respectively in an embodiment of a data processing system for facilitating AI-assisted verification of packaged and / or unpackaged drugs (such as Figure 1 Drug analysis engine server 155 and Fig.10 1000). Memory 1105 represents one or more storage devices containing software and data for facilitating the operation of the drug analysis engine server 155 and drug analysis engine module 160 described herein. Memory 1105 may include, but is not limited to, the following types of devices: cache, ROM, PROM, EPROM, EEPROM, flash memory, SRAM, and DRAM. Fig.11 As shown, the memory 1105 may contain three or more categories of software and / or data: an operating system 1110, a drug analysis engine module 1125, and a communication module 1140. Specifically, the operating system 1110 may manage the software and / or hardware resources of the data processing system and may coordinate the execution of programs by the processor. The drug analysis engine module 1125 may include an AI engine module 1130 and a consensus engine module 1135. The AI ​​engine module 1130 may be configured to execute the above-mentioned instructions for the machine learning engine 240, the convolutional neural network 310, and the Figure 7 The consensus engine module 1135 may be configured to perform the above operations for Figure 8 Consensus Engine 800 and Fig. 9The communication module 1140 may be configured to support communication between, for example, the drug analysis engine server 155 and, for example, a drug packaging authentication system.

[0077] although Figure 10-11 The present invention shows some embodiments of the present invention that can be used in a data processing system (such as Figure 1 Drug analysis engine server 155 and Fig.10 The hardware / software architecture of the data processing system 1000 may be described herein, but it will be understood that embodiments of the present invention are not limited to such configurations, but are intended to encompass any configuration capable of performing the operations described herein.

[0078] For ease of development, the programs used to execute the above programs can be written in a high-level programming language (such as Python, Java, C and / or C++). Figure 1-Figure 11 The computer program code for the operation of the data processing system discussed herein may be written in other programming languages ​​(such as, but not limited to, interpreted languages). Some modules or routines may be written in assembly language or even microcode to enhance performance and / or memory usage. It will also be understood that the functionality of any or all of the program modules may also be implemented using discrete hardware components, one or more application specific integrated circuits (ASICs), or programmed digital signal processors or microcontrollers.

[0079] In addition, according to various embodiments of the present invention, Figure 1 Drug analysis engine server 155 and Fig.10 The functions of the data processing system 1000 may be implemented as a single processor system, a multi-processor system, a multi-core processor system, or even a network of independent computer systems. Each of these processors / computer systems may be referred to as a "processor" or a "data processing system."

[0080] According to some embodiments of the inventive concepts described herein, the invention may be used Figure 1-Figure 11The data processing devices described herein are for facilitating verification of packaged and / or unpackaged pharmaceutical products. These devices may be embodied as operable to receive, send, process and store data using any suitable combination of software, firmware and / or hardware, and may be stand-alone, or interconnected via any public and / or private, real and / or virtual, wired and / or wireless network (including all or part of the global communications network known as the Internet) One or more enterprise, application, personal, ubiquitous and / or embedded computer systems and / or devices, and may include various types of tangible, non-transitory computer-readable media. Specifically, the memory 1105, when coupled to the processor, includes computer-readable program code that, when executed by the processor, causes the processor to perform operations including those described herein for Figure 1-Figure 9 Describes an operation that includes one or more operations.

[0081] As described above, embodiments of the present invention may provide an AI-assisted drug analysis system that may use one or more characteristics of a drug to pre-filter or pre-classify the drug into groups for training corresponding AI engines or models, rather than using a single larger AI engine or model trained on a comprehensive list of all drugs. Retraining a smaller AI engine or model when a new drug is introduced may be more efficient than retraining a single larger AI engine or model. In addition, some embodiments of the present invention may provide a consensus engine that may be based on rule-based decision trees and / or AI techniques to verify the accuracy of training information associated with a new drug by ensuring consistency between multiple sources of drug training information.

[0082] Other definitions and examples:

[0083] In the above description of various embodiments of the present disclosure, various aspects of the present disclosure may be shown and described herein in any form in a variety of patentable categories or contexts, including any new and useful process, machine, manufacture or composition of matter, or any new and useful improvement thereof. Therefore, various aspects of the present disclosure may be implemented entirely by hardware implementation, entirely by software (including firmware, resident software, microcode, etc.) implementation, or by combined software and hardware implementation, all of which may generally be referred to herein as "circuits", "modules" or "systems". In addition, various aspects of the present disclosure may take the form of a computer program product including one or more computer-readable media having computer-readable program code thereon.

[0084] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, (but not limited to) an electronic, magnetic, optical, electromagnetic or semiconductor system, device or equipment, or any suitable combination of the foregoing. More specific examples (non-exhaustive list) of computer-readable storage media will include the following: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a suitable optical fiber with a repeater, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer-readable storage medium may be any tangible medium that may contain or store a program used by an instruction execution system, device or device, or a program used in combination with an instruction execution system, device or device.

[0085] Computer readable signal media may include propagated data signals containing computer readable program code, such as in baseband or as part of a carrier wave. Such propagated signals may take any of a variety of forms, including but not limited to electromagnetic, optical, or any suitable combination thereof. Computer readable signal media may be any computer readable medium that is a non-computer readable storage medium and is capable of communicating, propagating, or transmitting a program for use by an instruction execution system, device, or equipment, or used in combination with an instruction execution system, device, or equipment. The program code embodied on a computer readable signal medium may be sent using any suitable medium (including but not limited to wireless, wired, optical fiber cable, RF, etc., or any suitable combination of the foregoing).

[0086] Computer program code for performing operations for various aspects of the present disclosure may be written in any combination of one or more programming languages ​​including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, conventional procedural programming languages ​​such as "C" programming languages, Visual Basic, Fortran 2003, Perl, COBOL 2002, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer may be connected to the user's computer through any type of network including a local area network (LAN) or a wide area network (WAN), or the connection may be connected to an external computer (e.g., through the Internet using an Internet service provider) or in a cloud computing environment, or provided as a service (such as software as a service (SaaS)).

[0087] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each frame in the flowchart and / or block diagram, and the combination of frames in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device for generating a machine, so that the instruction (executed via the processor of the computer or other programmable instruction execution device) creates a mechanism for implementing the function / action specified in one or more frames of the flowchart and / or block diagram.

[0088] These computer program instructions may also be stored in a computer-readable medium, which when executed may direct a computer, other programmable data processing device, or other device to operate in a particular manner, so that the instructions, when stored in the computer-readable medium, produce an article of manufacture, which includes instructions that, when executed, cause a computer to implement the functions / actions specified in one or more boxes of a flowchart and / or block diagram. The computer program instructions may also be loaded onto a computer, other programmable instruction execution device, or other device, so that a series of operating steps are executed on the computer, other programmable device, or other device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0089] The flow charts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the systems, methods and computer program products according to various aspects of the present disclosure. In this regard, each frame in the flow chart or block diagram can represent a module, segment or part of a code, which includes one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative embodiments, the functions marked in the frame may not occur in the order marked in the accompanying drawings. For example, the two frames shown in succession can actually be substantially concurrently executed, or these frames can be executed in reverse order at some point, depending on the functions involved. It will also be noted that each frame of the block diagram and / or flow chart, and the combination of the frames of the block diagram and / or flow chart can be implemented by a system based on hardware for a specific purpose that performs a specific function or action, or a combination of hardware and computer instructions for a specific purpose.

[0090] The terms used herein are only used for the purpose of describing specific aspects and are not intended to limit the present disclosure. As used herein, the singular forms "one", "an" and "said" are intended to also include plural forms, unless the context clearly indicates the opposite. It will also be understood that the terms "include", "comprise", "include", "contain", "have", "with" or their variations are used in this specification to indicate the presence of the features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof. As used herein, the term "and / or" includes any and all combinations of one or more items listed in association. Throughout the description of the accompanying drawings, similar reference numerals refer to similar elements.

[0091] It will also be understood that although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another.

[0092] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by a person of ordinary skill in the art to which the present invention belongs. It will also be understood that terms such as those defined in general dictionaries should be interpreted as having a meaning consistent with their meaning in the context of this specification and the relevant art, and should not be interpreted as an ideal or overly superficial meaning unless explicitly defined as such herein. For the sake of brevity and / or clarity, well-known functions or structures may not be described in detail.

[0093] The description of the present disclosure has been presented for the purpose of illustration and description, but is not intended to be exhaustive or limited to the disclosure in the form disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the present disclosure. Various aspects of the disclosure herein are selected and described to best explain the principles and practical application of the present disclosure, and to enable those of ordinary skill in the art to understand the present disclosure and various modifications suitable for the particular use contemplated.

Claims

1. A method comprising: receiving information associated with a drug, the information comprising a plurality of characteristics; filtering the information based on at least one of the plurality of characteristics to identify an artificial intelligence engine from a plurality of artificial intelligence engines; as well as Use one of the multiple artificial intelligence engines identified to predict the national drug code NDC of the drug, or use one of the multiple artificial intelligence engines identified to verify whether the drug matches the target drug.

2. The method according to claim 1, wherein: The plurality of artificial intelligence engines respectively correspond to a plurality of value combinations of at least one characteristic among the plurality of characteristics.

3. The method according to claim 2, wherein: At least one characteristic of the plurality of characteristics is not used to filter the information; and Among them, at least one characteristic among the multiple characteristics that is not used to filter the information is used as at least one feature for training the multiple artificial intelligence engines.

4. The method according to claim 1, wherein: The plurality of characteristics may include size, shape, color, imprint code or rating.

5. The method according to claim 4, wherein: The imprint code includes a marking of the pharmaceutical strength, a marking of the active ingredient, and a marking of the inactive ingredient.

6. The method according to claim 4, wherein: The shapes include circular, oval and other shapes.

7. The method according to claim 4, wherein: The colors include transparent and various colors.

8. The method according to claim 1, wherein: Filtering the information includes: The information is filtered based on all of the plurality of characteristics to identify an artificial intelligence engine from the plurality of artificial intelligence engines.

9. The method according to claim 1, wherein: Each of the plurality of artificial intelligence engines comprises a convolutional neural network.

10. The method according to claim 9, wherein: The convolutional neural network includes a plurality of convolutional layers, wherein at least some of the plurality of convolutional layers are connected to each other via skip connections.

11. The method according to claim 1, wherein: The plurality of artificial intelligence engines respectively correspond to a plurality of value combinations of at least one characteristic among the plurality of characteristics; The method further comprises: The entity is granted access to some of the multiple artificial intelligence engines based on the entity distributing some of the multiple drugs that respectively have some of the multiple value combinations corresponding to some of the multiple artificial intelligence engines.

12. A method comprising: receiving training information associated with a drug from a plurality of sources, the training information including a plurality of characteristics and a national drug code (NDC) of the drug; determining whether to accept, reject, or waitlist the training information associated with the drug based on consistency of the training information between some of the plurality of sources; as well as When the training information is received, the training information is used to train an artificial intelligence engine, wherein the artificial intelligence engine is configured to predict the NDC code of the drug based on the plurality of characteristics.

13. The method according to claim 12, wherein: Determining whether to accept, reject, or waitlist the training information includes: The training information is accepted when the training information is consistent between at least a consensus subset of some of the plurality of sources.

14. The method according to claim 13, wherein: The consensus subset includes a minimum number X of the plurality of sources.

15. The method according to claim 14, wherein: The consensus subset also includes a minimum percentage Y of the plurality of sources.

16. The method according to claim 12, wherein: Determining whether to accept, reject, or waitlist the training information includes: A consensus artificial intelligence engine is used to determine whether the training information is accepted, rejected, or waitlisted.

17. The method according to claim 16, wherein: The consensus AI engine uses K-means clustering to determine whether to accept, reject, or waitlist the training information.

18. The method according to claim 12, further comprising: Prior to accepting the training information, the training information is validated using a trusted source of the training information.

19. The method according to claim 12, wherein: The plurality of characteristics may include size, shape, color, imprint code or rating.

20. A system comprising: processor; as well as a memory coupled to the processor and comprising computer readable program code embodied in the memory and executable by the processor to perform operations comprising: receiving information associated with a drug, the information comprising a plurality of characteristics; filtering the information based on at least one of the plurality of characteristics to identify an artificial intelligence engine from a plurality of artificial intelligence engines; and Use one of the multiple artificial intelligence engines identified to predict the national drug code NDC of the drug, or use one of the multiple artificial intelligence engines identified to verify whether the drug matches the target drug.

21. A computer program product comprising: A non-transitory computer-readable storage medium including computer-readable program code embodied in the medium and executable by a processor to perform operations including: receiving information associated with a drug, the information comprising a plurality of characteristics; filtering the information based on at least one of the plurality of characteristics to identify an artificial intelligence engine from a plurality of artificial intelligence engines; and Use one of the multiple artificial intelligence engines identified to predict the national drug code NDC of the drug, or use one of the multiple artificial intelligence engines identified to verify whether the drug matches the target drug.

22. A system comprising: processor; as well as a memory coupled to the processor and comprising computer readable program code embodied in the memory and executable by the processor to perform operations comprising: receiving training information associated with a drug from a plurality of sources, the training information including a plurality of characteristics and a national drug code (NDC) of the drug; determining whether to accept, reject, or waitlist the training information associated with the drug based on consistency of the training information between some of the plurality of sources; and When the training information is received, the training information is used to train an artificial intelligence engine, wherein the artificial intelligence engine is configured to predict the NDC code of the drug based on the plurality of characteristics.

23. A computer program product comprising: A non-transitory computer-readable storage medium including computer-readable program code embodied in the medium and executable by a processor to perform operations including: receiving training information associated with a drug from a plurality of sources, the training information including a plurality of characteristics and a national drug code (NDC) of the drug; determining whether to accept, reject, or waitlist the training information associated with the drug based on consistency of the training information between some of the plurality of sources; and When the training information is received, the training information is used to train an artificial intelligence engine, wherein the artificial intelligence engine is configured to predict the NDC code of the drug based on the plurality of characteristics.

Images